Maybe one step towards this would be to create a benchmark that measures how much a models knows about alignment.
Yes! This would be valuable. Generally, getting a sense of the ‘self-awareness’ of a model in terms of how much it knows about itself would be a valuable thing to start testing for.
Maybe one step towards this would be to create a benchmark that measures how much a models knows about alignment.
Yes! This would be valuable. Generally, getting a sense of the ‘self-awareness’ of a model in terms of how much it knows about itself would be a valuable thing to start testing for.